Artificial Intelligence (AI) is already widely used in everyday communication. Despite concerns about the negative impacts AI may have on society, the social effects of using AI in communication have been largely understudied. This research investigates the social impact of algorithm-based reply suggestions, known as “smart replies,” which are one of the most prevalent AI applications and are used to send billions of messages daily.
Through two randomized experiments, it was demonstrated that recommendation systems based on such algorithms can alter people’s interactions and perceptions in both prosocial and antisocial ways. The study findings reveal that the use of algorithmic replies can alter language and social relationships. Specifically, it was found that:
- Communication speed increases.
- Use of positive emotional expressions rises.
- Conversation partners are perceived as more intimate and cooperative.
However, in line with general assumptions about the negative effects of AI, individuals suspected of using algorithmic replies are more likely to be evaluated negatively.
Therefore, while AI has the potential to enhance communication speed and improve interpersonal perception, the general antisocial impression of AI may undermine these potential benefits, especially when the use of AI is evident.
This research sheds light on the complex social impacts of AI-mediated communication, emphasizing the importance of balancing the advantages and disadvantages of AI.